Truace
Subscribe

FAHR-Face AI model for predicting cancer patient survival from facial images

Source article: Can AI really predict cancer survival better than a doctor? I have my doubts | Devi Sridhar

The idea that a single selfie could predict how long someone with cancer may live sounds like an incredible breakthrough. AI is getting better at spotting patterns in medical data, but can it be used to reliably predict what will happen to an individual patient? I asked a clinical colleague what she thought about the recent headlines. She found it an interesting development given that doctors have long used facial appearance to assess patient health. For example, she noted that they may write “appears stated age…

By the Truvace desk2 min read1 sourceJournalismTRV-2026-2219· cite & verify
Can AI really predict cancer survival better than a doctor? I have my doubts | Devi Sridhar
Trace impact readingContested
P 59The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 56The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

Researchers at Mass General Brigham built FAHR-Face, a model trained on more than 40 million facial images, to estimate biological age and predict survival among cancer patients. A survival-specific version used almost 35,000 photos of patients, and results appeared in The Lancet Digital Health.

The commentary argues the work is promising but unproven for everyday care. Open questions include whether it beats clinicians' own judgment, whether it holds across hospitals and patient groups, and whether photo variation and cosmetic changes undermine the estimates.

Main points

  1. The survival model focused on areas below the eyes and the bridge of the nose, regions the lead author linked to signs described by Hippocrates.
  2. Facial age was significantly linked to survival in 18 of 28 cancer categories, so it works for some cancers but not others.
  3. The measure was not significantly associated with survival for Black patients after adjustment, raising concern about racial bias in training data.
  4. The author notes lighting, angle, expression, botox, fillers, facelifts and GLP-1 drugs can alter facial appearance and age estimates.

The gain

An AI model trained on facial images found AI-estimated facial age significantly associated with survival in 18 of 28 cancer categories, suggesting it could add prognostic information beyond a doctor's visual impression.

The problem

The facial-age measure was not significantly associated with survival for Black patients after adjustment, and selfie variability, cosmetic changes and retrospective design limit reliable individual prediction.

The rundown

The study created two tools: FAHR-FaceAge, which estimates age from features such as the forehead, wrinkles between nose and mouth, and temples, and FAHR-FaceSurvival, trained on cancer patients' photos. A clinical colleague noted that doctors already use phrases like appears stated age as a rough health proxy.

The author stresses that the models rest on retrospective statistical associations, and moving from that to accurately predicting an individual's future survival is a jump. Researchers say they tested the age model against lighting, expression and cosmetic surgery, but the author doubts error can be fully avoided.

Prof Miroslaw Bober said the tool still needs testing against doctors' judgments and across hospitals and patient groups. The public website also adds uploaded selfies to the study under its consent forms, and currently reports only biological age.

What this doesn’t fix

The tool has not been tested against doctors' own judgments or across different hospitals and patient groups, and the models rely on retrospective associations rather than proven clinical usefulness.

Sources

  1. JournalismThe Guardian2026-10-23

The debate